From: julie langou
Date: Tue, 25 Oct 2011 09:02:54 +0200
This has been corrected in rev 1028.
The tests for n=0 have been skipped.
Thank you for the report.
On Oct 22, 2011, at 9:37 PM, Ake Sandgren <ake.sandgren@Domain.Removed> wrote:
NORMA = SLANSY( NORM, UPLO, N, A, LDA, WORK )
RESULT(1) = ( NORMA - NORMARF ) / NORMA / EPS
N = 0 -> slansy = 0 -> norma = 0 -> div by zero
not protected by IEEE checking.
Lapack mailing list